Writing · 22 posts · 7 years
Blog.
Notes on distributed training, ML infrastructure, Kubernetes runbooks, and personal essays.
Featured
01
Rhumbatron: Many AI Agents on One Codebase
How we built a multi-agent Git platform on Cloudflare Workers, Artifacts, and Durable Objects that detects semantic conflicts, verifies candidate builds in sandboxes, and promotes only passing releases.
#ai #ai-agents #software-engineering
02
PyTorch DDP for Fine-Tuning Gemma 4 on Multiple GPUs
How PyTorch DistributedDataParallel works and when it is the right tool for LoRA fine-tuning Gemma 4, with a complete torchrun script, memory math for the 12B model, and the failure modes that hang multi-GPU jobs.
#pytorch #machine-learning #ai
03
How I Consolidated Milvus Across AZs
Step-by-step runbook for migrating a multi-AZ Milvus deployment onto a single dedicated Kubernetes node without losing a single vector.
#aws #devops #kubernetes
2026
5 posts
-
Rhumbatron: Many AI Agents on One Codebase
How we built a multi-agent Git platform on Cloudflare Workers, Artifacts, and Durable Objects that detects semantic conflicts, verifies candidate builds in sandboxes, and promotes only passing releases. #ai #ai-agents #software-engineering -
Real-Time Transformer Prediction at Autodesk
Notes from my summer 2022 ML internship at Autodesk in San Francisco, where I shipped a low-latency transformer into a live event pipeline scoring 1M+ daily user interactions. #machine-learning #ai #portfolio -
PyTorch DDP for Fine-Tuning Gemma 4 on Multiple GPUs
How PyTorch DistributedDataParallel works and when it is the right tool for LoRA fine-tuning Gemma 4, with a complete torchrun script, memory math for the 12B model, and the failure modes that hang multi-GPU jobs. #pytorch #machine-learning #ai -
Making Your Site Agent Friendly
A standards-based guide to make a website discoverable, readable, callable, trustworthy, and verifiable by AI agents. #ai #software-engineering #devops -
K3s on Oracle Cloud Always Free: GitOps Kubernetes (Gateway API + Auto HTTPS)
Running a 3-node K3s cluster on Oracle Cloud Always Free with Terraform, Argo CD GitOps, Envoy Gateway (Gateway API), and automated Let’s Encrypt TLS. #kubernetes #devops #oracle-cloud
2025
2 posts
-
Bare Metal Kubernetes Homelab Setup
A self-hosted bare-metal Kubernetes platform on Ubuntu 24.04 LTS, provisioned with Ansible and reconciled via ArgoCD GitOps. #devops #kubernetes #software-engineering -
How I Consolidated Milvus Across AZs
Step-by-step runbook for migrating a multi-AZ Milvus deployment onto a single dedicated Kubernetes node without losing a single vector. #aws #devops #kubernetes
2024
2 posts
-
conda vs poetry vs uv vs pip
How I combine conda, poetry, uv, and pip to keep local Python environments sane and make CI/Docker builds 10–100× faster. #python #devops -
Chat with SQLite Database
A lightweight Python agent built with LangChain and OpenAI function calling that translates natural-language questions into SQLite queries and generates HTML reports. #portfolio #ai #openai
2023
4 posts
-
LangChain Chat and Chroma Embeddings Integration
A two-part Python sandbox exploring stateful terminal chat with ConversationSummaryMemory and local retrieval-augmented generation (RAG) using Chroma. #ai #openai #portfolio -
Generic Code Generator
PyCode is a sequential LangChain CLI utility that generates implementation code and matching unit tests in any target programming language. #ai #openai #portfolio -
AI Free Will: The Ethics and Implications of Autonomous Machines
Can an autonomous machine ever choose freely? An essay on determinism, moral responsibility, and what happens to law and ethics when machines act on their own. #ai -
The Struggle With AI Generated Art
Reflecting on the tension between diffusion models like DALL-E and Stable Diffusion and the working artists whose styles trained them. #ai #machine-learning
2022
1 post
2021
6 posts
-
Install TensorFlow on Apple M1 (M1, Pro, Max) with GPU (Metal)
How to install native arm64 TensorFlow with the tensorflow-metal GPU plugin on Apple M1, M1 Pro, and M1 Max Macs using Miniforge. #getting-started #machine-learning #tensorflow -
Install Kubeflow on Digital Ocean Managed Kubernetes (with Terraform)
A practical guide to provisioning a DigitalOcean Managed Kubernetes cluster with Terraform and deploying Kubeflow 1.2 with doctl, kubectl, and kfctl. #docker #kubernetes #digital-ocean -
Data Analysis for a Food Tech Startup: Swiggy
Exploratory analysis and customer segmentation on a 10,000-row Swiggy order sample using RFM quartiles, custom delivery-efficiency scoring, and k-means clustering. #portfolio -
CoGeoTIFF Research on Space Tech SaaS Platform at Pixxel
Engineering research memo from my time at Pixxel evaluating Cloud Optimized GeoTIFFs (COGs), dynamic tile servers, Sentinel-2 conversion pipelines, and AWS unit economics. #portfolio -
Leaving Pixxel
Wrapping up an unforgettable year at Pixxel, and a thank-you note to the people who built satellites and geospatial pipelines with me. #career #goodbye -
Install JupyterHub on AWS Elastic Kubernetes Service (EKS)
Field notes and step-by-step runbook for deploying multi-user, GPU-enabled JupyterHub on AWS EKS using custom Docker images and Zero to JupyterHub. #aws #kubernetes #machine-learning
2020
2 posts
-
Machine Learning Best Practices For Your Company
An onboarding checklist for ML and data engineering teams on Git hygiene, secret handling, DVC, notebook discipline, and experiment tracking. #machine-learning #devops -
Resume Tips That Will Help You Land Your Software Engineering Job
Practical, no-nonsense advice for writing a software engineering resume: what recruiters actually scan for, what to cut, and how to frame your impact. #career